Summary
One call sits between a snippet and the answer to what it actually prints.
The parameters are the code and its languageId — there is no session id, no working directory, no file handle — so each call stands alone and nothing you define in one is waiting in the next. That makes it a scratchpad for settling a question about behaviour, not a place to build something up across several steps.
What it is
A single-tool server that runs a code snippet in a named language and returns the result.
What you get
- A snippet executed and its result returned to the conversation
- The language chosen explicitly through languageId, rather than guessed from the code
- One entry point to reason about: what goes in is code, what comes back is what it printed
Requirements
Node to run the server under, the snippet itself, and a languageId naming its language.
Setup effort
One command — npx -y mcp-server-code-runner